588 Prof. 0. W. Richardson and Mr. K. T. Compton on 



It is thus about 20 per cent, less than the theoretical value. 

 The slope of OD is almost exactly half of that of OB. The 

 slope of OC is only about two thirds of that of OB, but the 

 same relation holds between OE and 00 as between OD and 

 OB. It thus appears that s is very close to one half for all 

 the metals investigated, even including copper and bismuth, 

 for which the discrepancy between k m and h is much greater 

 than in the case of the other six metals. 



The relation T r =sh(v — v ) has only been shown theo- 

 retically to hold for the case of isotropic radiation, that is for 

 incident radiation which is propagated with equal intensity 

 in all directions, whereas all the values given are for light at 

 approximately normal incidence. However, we made special 

 experiments with platinum and aluminium to see whether 

 rotating the strip S so as to change the angle of incidence of 

 the light would make any alteration in the curves giving the 

 distribution of kinetic energy. We were unable to detect 

 any certain changes in the energy in this way, although the 

 number of electrons emitted varied considerably. It would 

 appear, therefore, that the difference in the distribution of 

 energy among the electrons emitted by isotropic radiation 

 and those emitted by radiation incident at a particular angle, 

 is inappreciable or, at least, its determination is not a 

 practical matter when the degree of precision which is at 

 present attainable in experiments of this kind is taken into 

 consideration. 



The fact that all of the measured values of XT m fall to the 

 left of the line OA is perhaps not very surprising. Prac- 

 tically every source of experimental error tends to make the 

 measured values of T m too small. This would be the case, 

 for example, if the surface of the strip were covered with a 

 layer of photoelectrically inactive material such as the oxide 

 of the metal or a layer of condensed gas or moisture ; the 

 effect of any slight leakage in the electrostatic system is also 

 in the same direction. We are unable at present to urge 

 any satisfactory reason why the metals copper and bismuth 

 should appear to be in a class by themselves. 



It is worth while remarking that the difference between 

 the experimental and the theoretical relation of T m to v is 

 what one would expect if there were a layer of photo- 

 electrically inactive material at the surface of the strips 

 which reduced the speed of the escaping electrons, provided 

 the law of diminution of kinetic energy T with thickness x of 

 material traversed were of the form 



ST=-T/(^)Stf (14) 



